A paralyzed man imagined writing with a hand that could no longer move, and a computer turned those intended movements into words on a screen in real time. The result was not science fiction: in a 2021 Nature study, an implanted brain-computer interface decoded attempted handwriting at about 90 characters per minute, with 94.1 percent raw online accuracy. Four years later, researchers went further, demonstrating that neural activity associated with silent inner speech could also be decoded in people with severe speech and motor impairments. The machine was no longer waiting for a person to move a cursor or speak aloud. It was beginning to interpret signals associated with words that existed only as thought.
That does not mean corporations can currently sit across a room and read your private thoughts. They cannot. Today’s systems require highly controlled conditions, specialized hardware, trained algorithms, and—at the cutting edge—implanted electrodes. But the direction of travel is unmistakable. Neurotechnology is moving from measuring the brain to decoding useful information from it, while researchers, governments, courts, ethicists, and companies are confronting a question previous generations never had to formulate: what happens when information generated inside a person’s brain becomes technologically accessible to someone else?
The Machine That Learned to Read Handwriting Before the Hand Moved
The 2021 Stanford experiment is a better place to begin than almost any futuristic prediction. The participant, a man with paralysis caused by spinal-cord injury, was asked to imagine writing letters with a pen. Electrodes implanted in the motor cortex recorded the neural activity associated with those attempted movements, and a recurrent neural network translated the signals into text. He reached approximately 90 characters per minute with 94.1 percent raw accuracy during online use. With a general-purpose autocorrect system applied offline, accuracy exceeded 99 percent. The significance was not that a computer had somehow extracted an unrestricted stream of consciousness. It was that a specific, highly structured intention—attempting to write a particular character—could be decoded from neural activity rapidly enough to become practical communication.
Then the Research Moved Closer to Inner Speech
In August 2025, Stanford researchers reported another step that deserves considerably more attention than the sensational phrase “mind reading” usually receives. Working with four people with severe speech and motor impairments who had microelectrode arrays implanted in motor areas of the brain, the researchers found robust neural representations associated with inner speech and demonstrated real-time decoding of imagined sentences. The study also explored free-form inner speech during tasks such as sequence recall and counting. Most importantly, the researchers did not present this as unrestricted access to a person’s private mind: the work was a proof of principle performed under experimental conditions, and the signals associated with inner speech were closely related to those generated during attempted speech. The researchers also demonstrated strategies designed to prevent a speech BCI from unintentionally decoding private inner speech.
That distinction is crucial. “A machine can decode some neural signals under controlled experimental conditions” and “a machine can secretly read everything you think” are not remotely equivalent statements. The first is now demonstrated science. The second remains science fiction. But the distance between those propositions is precisely what makes the first scientifically important: every new capability changes the boundary of what future systems might eventually be able to infer.
What the Brain Signal Contains—and What It Doesn’t
A neural recording is not a magical transcript of consciousness. It is a stream of biological signals whose meaning depends on where they were recorded, how the experiment was structured, what the participant was doing, how much training the decoder received, and what the algorithm was specifically designed to recognize. Today’s successful demonstrations generally involve participants performing constrained tasks with implanted electrodes, rather than an observer passively extracting arbitrary memories, beliefs, fantasies, passwords, or private conversations from an unsuspecting person’s head.
That limitation should make the issue more precise, not less urgent. The privacy problem begins before perfect mind reading. If a device can extract information about intended movement, attempted speech, inner speech, attention, or other neurological states, then neural data can become extraordinarily sensitive information even when the decoder remains imperfect. The question is therefore not simply whether a company can read everything. It is whether people should have meaningful control over increasingly intimate information that can be inferred from their nervous system.
The Neuro-Rights Debate Is Already Real
This is not a hypothetical ethical movement waiting for futuristic technology to arrive. Scientists and policymakers have been debating neurotechnology governance for years. In 2017, neuroscientist Rafael Yuste convened the Morningside Group, bringing together neuroscientists, ethicists, engineers, and clinicians to examine the social implications of rapidly advancing neurotechnology. The group later published recommendations addressing privacy, identity, agency, bias, and enhancement. The resulting “neurorights” movement argues that existing human-rights principles may need to be clarified or strengthened as technology becomes capable of recording, interpreting, or altering neural activity.
The terminology varies, and that matters. The Neurorights Foundation advocates five categories: mental privacy, personal identity, free will or agency, fair access to mental augmentation, and protection from algorithmic bias. But even the Foundation emphasizes that the project is not necessarily about inventing five entirely new human rights from scratch, it is also about interpreting and adapting existing rights to a technology capable of creating unprecedented forms of access to the brain.
Chile Put Brain Activity Into Constitutional Law
Chile became the most visible example of what this debate can produce at the national level. Its constitutional reform, Ley N° 21.383, published in the Diario Oficial on October 25, 2021, amended Article 19, Number 1 of the Constitution, adding protection relating to brain activity and information derived from it, the world’s first constitutional protection of its kind. The move became a landmark in the global neurorights debate because it treated neurotechnology not merely as another consumer-privacy problem but as something potentially connected to constitutional protection of the person.
But Chile should not be presented as proof that the world has already created a universal legal category called “mind ownership.” It has not. Different jurisdictions approach the problem through different combinations of constitutional rights, medical law, privacy law, data protection, consumer protection, bioethics, and human-rights principles. The unresolved question is precisely how far those existing protections reach when the data being collected originates directly from neural activity.
This Is Already an International Policy Problem
The international response has moved well beyond academic discussion. The OECD adopted a Recommendation on Responsible Innovation in Neurotechnology in 2019, calling for safety assessment, inclusivity, scientific collaboration, societal deliberation, oversight, protection of personal brain data, stewardship, and anticipation of misuse. The Council of Europe subsequently examined whether its existing human-rights framework is sufficient to address neurotechnology, including questions surrounding privacy, personhood, discrimination, cognitive liberty, mental privacy, and mental integrity. In 2026, the Council of Europe published a dedicated report from a workshop on human rights and neurotechnologies, while its broader 2026–2030 strategy continues to treat emerging biomedical technologies as a forward-looking human-rights governance issue.
The United Nations has entered the discussion as well. The Human Rights Council has examined the impact, opportunities, and challenges of neurotechnology for the promotion and protection of human rights, with particular attention to questions that become difficult when private companies, rather than governments, control the relevant technologies and data. The important development is not that the international community has agreed on a single new “right to your thoughts.” It is that major institutions are now treating neural data and neurotechnology as a legitimate human-rights and governance problem.
The Commercial Question Is Bigger Than Brain Implants
There is an easy mistake in this discussion: imagining that the only relevant technology is an implanted chip like the systems used in clinical trials. It isn’t. Neurotechnology includes a much wider family of devices and methods capable of recording, influencing, or analyzing activity in the brain and nervous system. Some applications are medical. Others are increasingly consumer-oriented. The more neural information becomes commercially valuable, the more important it becomes to ask what happens to the data after it leaves the device.
A conventional password is information you can change. A credit-card number can be replaced. A location history can sometimes be deleted or reset. Your neural activity is different. The underlying biological system generating it is part of you. That does not automatically make every neural signal “your property” in a simple legal sense, but it does make the consequences of misuse unusually intimate. A company that mishandles a shopping history may expose what you bought. A system that mishandles neural data could potentially expose information about how your brain responds, what signals you generate under particular conditions, or what the algorithm believes those signals mean.
From Clicks to Attention—and Eventually to Intention
There is a broader technological trajectory worth watching, but it should be described without turning it into prophecy. The commercial internet learned to monetize clicks and browsing behavior. Recommendation systems learned to infer preferences from patterns of activity. Modern AI systems can infer likely intentions from language and context. Neurotechnology introduces another possible layer: extracting useful information directly from biological signals associated with action, speech, or thought.
That does not establish a straight line from targeted advertising to mind control. Human brains are not browser histories, and neural signals are vastly more difficult to interpret than clicks. But the economic incentive is easy to understand. The closer technology gets to information a person cannot simply observe from the outside, the more valuable control over that information could become. The governance problem therefore arrives before the science-fiction endpoint. Society does not have to wait until a machine can read an entire novel from someone’s mind before deciding that neural information deserves unusually strong protection.
What Dream Research Actually Shows
Dream research provides another useful reality check because it demonstrates that technology can influence aspects of subjective experience without remotely approaching the fantasy of uploading advertising directly into a sleeping mind. At MIT, researchers developed Dormio and a method called Targeted Dream Incubation, using a wearable system and carefully timed auditory cues around sleep onset to steer dream content toward particular themes. Peer-reviewed work has reported increased creative performance following targeted dream-incubation sessions. This is real experimental science—but it is controlled dream induction, not commercial mind control.
That distinction matters because the future risk is often easiest to understand by looking at what the technology can actually do today. Researchers can influence sleep-onset experience under experimental conditions. They can decode specific categories of neural activity. They can turn attempted handwriting into text. They can extract limited information associated with inner speech. None of those achievements establishes that advertisers can secretly plant branded dreams or read unrestricted thought. They do establish something more consequential: the brain is becoming an increasingly accessible technological interface.
The Most Important Word Is Consent
The deepest issue may therefore not be ownership at all. It may be consent.
A person with paralysis may willingly undergo an invasive procedure because the ability to communicate is worth the medical risk. A patient may agree to neural recording as part of a clinical trial. A consumer may agree to use a headset for a particular application. But consent becomes much harder to define when the same underlying signal can potentially be processed for purposes the person did not anticipate. Can consent to operate a device also become consent to collect neural data indefinitely? Can data gathered for medical rehabilitation later be used for research, advertising, insurance, employment, or algorithmic profiling? How long should it be retained? Can it be sold? Can it be subpoenaed? Can a person demand its deletion? Can an algorithmic inference about a person’s mental state be treated as fact when the underlying neural signal is probabilistic and context-dependent?
The Future Fight May Not Be About Reading Thoughts
The most important misconception to eliminate is that we need perfect mind reading before this becomes a serious political problem. We don’t. A technology can be imperfect and still be commercially valuable. A neural decoder that is wrong often may still reveal enough information to create privacy, discrimination, or autonomy concerns. An algorithm does not need to know everything about you to make a consequential prediction about you.
That is why the debate is moving now. The OECD is already calling for protection of brain data. The Council of Europe is examining whether existing rights adequately cover neurotechnology. The United Nations has studied its human-rights implications. Researchers are explicitly testing whether private inner speech can be decoded and, importantly, how to prevent unintended decoding. Consumer-neurotechnology companies are already being examined for their data practices. The law is being forced to confront a category of information that previous privacy regimes were never designed around.
So Who Owns Your Thoughts?
The honest answer is that there is no single worldwide legal rule saying that you “own your thoughts” as a distinct category of property. And that may be exactly the wrong framework.
The more fundamental principle may be that some forms of information are so intimately connected to personhood that ordinary data-ownership rules are inadequate. Mental privacy, cognitive liberty, bodily integrity, identity, agency, informed consent, and freedom of thought may ultimately matter more than the familiar language of who owns a database record. The question is not merely whether a company owns the server on which neural data is stored. It is whether the person whose brain produced that data retains meaningful power over what the technology is allowed to infer, remember, transmit, and do with it.
That is the real neuro-rights problem.
We are not yet living in a world where corporations can casually read the private contents of everyone’s minds. The science is nowhere near that. But we are already living in a world where a machine can translate attempted handwriting into language, where researchers can decode limited forms of inner speech, where dreams can be experimentally nudged toward particular themes, and where governments and international institutions are beginning to ask whether the rights designed to protect human beings are sufficient for a technology that can reach inside the biological machinery of thought.
The decisive battle will not begin when a machine finally reads a person’s entire mind. By then, the rules will already have been written—or failed to be written. The real question is being decided much earlier, while the technology is still imperfect enough to look harmless: when the boundary between a person’s brain and a machine becomes technically permeable, does the law protect the machine’s access—or the person’s mind?
